On \(k\)-major norms and \(k\)-minor antinorms (Q1200575): Difference between revisions

The \(k\)-major norm \(| X|_{[k]}\) of a complex \(n\)-vector \(X\) is defined to be the sum of the \(k\) largest moduli of its components. If \(A\) is a complex \(m\times n\) matrix with columns \(A_ i\), the \(k_ 1,k_ 2\)-major norm of \(A\) is the \(k_ 2\)-major norm of the vector \(\bigl(| A_ 1|_{[k_ 1]},\dots,| A_ n|_{[k_ 1]}\bigr)\). The \(k_ 1,k_ 2\)-major operator norm of \(A\) is the max of all \(| AX|_{[k_ 1]}\), where \(X\) is a complex \(n\)-vector and \(| X|_{[k_ 2]}=1\). For real nonnegative \(X\) and \(A\), the \(k\)-minor norm of \(X\) is the sum of the \(k\) smallest moduli of its components and definitions analogous to those above are given for the \(k_ 1,k_ 2\)-minor antinorm of \(A\) and the \(k_ 1,k_ 2\)-minor operator antinorm of \(A\). The sub (super) multiplicativity of these matrix norms (antinorms) are characterized. New lower and upper bounds for the Perron root of nonnegative \(A\) are obtained. Examples are given in which some of these bounds are better than the Frobenius bounds and/or bounds due to \textit{E. Deutsch} [Pac. J. Math. 92, 49-56 (1981; Zbl 0478.15014)] and \textit{T. Szulc} [Linear Algebra Appl. 101, 181-186 (1988; Zbl 0659.15015)].